CN105759267A - Improved Omega-K imaging method of large squint SAR - Google Patents

Abstract

The invention provides an improved Omega-K imaging method of a large squint SAR (synthetic aperture radar), comprising: successively performing range direction rapid Fourier transform, range direction pulse compression, range direction inverse discrete Fourier transform, motion compensation, two-dimensional rapid Fourier transform and consistent compression on original echo data S; calculating frequency spectrum offset f'Tau,min to correct a Stolt interpolation; performing range direction inverse discrete Fourier transform on each data point; and performing linear phase compensation and azimuth inverse discrete Fourier transform to obtain a final imaging result. The method can save hardware storage resources, and guarantee imaging quality, is simple to operate, and has high algorithm efficiency.

Description

A kind of improvement Omega-K formation method of large slanting view angle machine SAR

Technical field

The invention belongs to SAR technical field of imaging, the improvement Omega-K especially relating to a kind of large slanting view angle machine SAR becomes Image space method.

Background technology

Omega-K algorithm is a kind of classical one-tenth of synthetic aperture radar (synthetic aperture radar is called for short SAR) As algorithm, it is by unanimously having compressed being fully focused at reference distance in two-dimensional frequency, then passes through Stolt Interpolation is without being nearly completed the remaining range migration correction (RCMC) of non-reference distance, remaining secondary range compression (SRC) And remnants Azimuth Compression.The mapping relations of Stolt interpolation are:

$\sqrt{{(f_0+f_{\mathrm{\τ}})}^2-\frac{c^2{f}_{\mathrm{\η}}^2}{4V_r^2}}=f_0+f_{\mathrm{\τ}}^{\′}---\left(1\right)$

Wherein, f₀For carrier frequency, f_τFor frequency of distance, c is the light velocity, f_ηFor orientation frequency, V_rFor radar speed, f_τ' it is the frequency of distance after mapping.Above formula is by original frequency of distance f_τIt is mapped as new frequency of distance f_τ', residual phase It is f_τ' linear function, thus eliminate residual phase modulation, it is achieved that the vernier focusing of the target of non-reference position.

Stolt mapping can cause displacement and the distortion of frequency spectrum, and the biggest this phenomenon in angle of strabismus is the most obvious.It is more than in angle of strabismus During certain value, after Stolt mapping, displacement and the distortion of frequency spectrum can cause spectrum component to lose, sternly beyond the scope of support region Important place have impact on image quality.

In the past by using extension Omega-k algorithm, considerably increasing the computational complexity of Stolt interpolation, real-time is poor； And by the method expanding two dimension support region in Stolt interpolation, then exchange image quality for sacrifice hardware store resource, Today that SAR echo data is huge, efficiency of algorithm certainly will be made a big impact.

Summary of the invention

Technical problem solved by the invention is to provide the improvement Omega-K formation method of a kind of large slanting view angle machine SAR, By computed range to frequency spectrum through Stolt map after side-play amount, re-define interpolation front distance frequency map f_τ' scope, Revise Stolt interpolation so that two-dimensional frequency falls in former support region, thus save hardware store resource, it is ensured that imaging Quality, improves efficiency of algorithm.

The technical solution realizing the object of the invention is:

The improvement Omega-K formation method of a kind of large slanting view angle machine SAR, comprises the following steps:

Step 1: obtain raw radar data S；

Step 2: raw radar data S is successively carried out distance to fast Fourier transform FFT, distance to pulse compression, Distance is compressed with consistent to inverse discrete Fourier transform IFFT, motion compensation, Two-dimensional FFT；

Step 3: calculate spectrum offset amount f_τ'_,min, revise Stolt interpolation；

Step 4: each data point is carried out distance to IFFT, transform data to range-Dopler domain, obtain data point S_ik；

Step 5: according to f_τ'_,minCarry out linear phase compensation, complete distance to Spectrum Correction, obtain data point S'_ik；

Step 6: to data point S'_ikCarry out orientation to IFFT, obtain final imaging results.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, original echo in step 1 The size of data S is Na × Nr, wherein, Na be orientation to sampling number, Nr is that distance is to sampling number.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, through consistent in step 2 Data after compression store with the form of two-dimensional matrix.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, step 3 specifically include with Lower step:

Step 3-1: computer azimuth is to frequency of distance f of unit_τIt is mapped to f by Stolt_τ' the minima of axle, quantify to take After whole, f is arrived in storage_τ'_,minIn；

Step 3-2: with f_τ'_,minAs initial value, withFor frequency interval, the frequency of distance calculating each data point is reflected Penetrate f_τ' value；

Step 3-3: by Stolt mapping equation, calculate the f of each data point_τ' it is worth correspondence at f_τThe position of axle, and count Calculate interpolation result.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, i-th in step 3-1 Orientation is to frequency of distance f of unit_τIt is mapped to f by Stolt_τ' the computational methods of minima that quantify after rounding of axle are:

$f_{\mathrm{\τ},min}^{\′}\[i\]=ceil\left(sqrt\right({\left(f_0+f_{\mathrm{\τ}}\[0\]\right)}^2-\frac{c^2{f}_{\mathrm{\η}}^2\[i\]}{4V_r^2})/(f_s/Nr\left)\right)$

Wherein, f_τ'_,min[i] represents the i-th orientation f to unit_τ' quantify the minima after rounding, f_τ[0] distance frequency is represented Rate initial value, f_η[i] represent i-th orientation to the orientation of unit to frequency, f_sRepresenting sample frequency, Nr is original time The distance of wave datum S is to sampling number, f₀Representing carrier frequency, c represents the light velocity, V_rRepresent radar speed.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, i-th in step 3-3 Orientation is to the f of unit_τ' it is worth correspondence at f_τThe position of axle is:

$f_{\mathrm{\τ},ik}=\sqrt{{(f_0+f_{\mathrm{\τ},ik}^{\′})}^2+\frac{c^2{f}_{\mathrm{\η}}^2}{4V_r^2}}-f_0$

Wherein, i be orientation to coordinate, k is that distance is to coordinate, f_τ'_,ikRepresent that position coordinates is (i, the distance of data point k) Frequency is mapped in f_τ' the value of axle, f_τ,ikRepresent f_τ'_,ikCorresponding at f_τThe position of axle.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, step 3-3 uses sinc Interpolation calculates interpolation result.

Further, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention, linear phase in step 5 Data point after compensation is:

$S_{ik}^{\′}=S_{ik}\exp \left(j2\mathrm{\π}k\frac{f_{\mathrm{\τ},min}^{\′}\[i\]}{Nr}\right)$

Wherein, f_τ'_,min[i] represents the i-th orientation f to unit_τ' quantifying the minima after rounding, Nr is original echo number According to the distance of S to sampling number, k is positive integer.

The present invention uses above technical scheme compared with prior art, has following technical effect that

1, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention need not expand support region, saves hardware Storage resource；

2, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention is prevented from Stolt interpolation and makes frequency spectrum oblique Tension-torsion song exceeds support region, fully possesses all of spectrum component；

3, the improvement Omega-K formation method of the large slanting view angle machine SAR of the present invention is while ensureing image quality, computing Simply, improve efficiency of algorithm.

Accompanying drawing explanation

Fig. 1 is the improvement Omega-K formation method flow chart of the large slanting view angle machine SAR of the present invention；

Fig. 2 is the frequency spectrum before and after tradition Omega-K algorithm Stolt interpolation, the frequency spectrum before wherein (a) is Stolt interpolation, B () is the frequency spectrum after Stolt interpolation；

Fig. 3 is the scattergram of the emulation experiment point target of the present invention；

Fig. 4 is the frequency spectrum in the improvement Omega-K algorithm of the present invention, and wherein (a) is the frequency spectrum before revising Stolt interpolation, B () is the frequency spectrum after correction Stolt interpolation, (c) is the frequency spectrum after linear phase compensates；

Fig. 5 is the image of the emulation experiment of the present invention；

Fig. 6 is the contour map of the present invention, wherein the contour map of point target centered by (a), and (b) is upper right point mesh Target contour map.

Detailed description of the invention

Embodiments of the present invention are described below in detail, and the example of described embodiment is shown in the drawings, the most extremely Same or similar label represents same or similar element or has the element of same or like function eventually.Below by ginseng The embodiment examining accompanying drawing description is exemplary, is only used for explaining the present invention, and is not construed as limiting the claims.

The flow chart of the improvement Omega-K formation method of a kind of large slanting view angle machine SAR that the present invention proposes is as shown in Figure 1. Mainly include pulse compression, motion compensation, two-dimensional fast fourier transform FFT, consistent compression, revise Stolt interpolation, Distance to inverse discrete Fourier transform IFFT, linear phase compensates, orientation is to IFFT.With traditional Omega-K algorithm Difference be to compensate and instead of tradition Stolt interpolation with revising Stolt interpolation and linear phase.Below will be from signal The two step is further explained by the angle processed.

First correction Stolt is illustrated.

Step one, computer azimuth are to frequency of distance f of unit_τIt is mapped to f by Stolt_τ' the minima of axle, quantify to round Rear storage is to f_τ'_,minIn.

After consistent compression, the remaining phase theta of 2-d spectrum_REF(f_τ,f_η) be approximately:

${\mathrm{\θ}}_{REF}(f_{\mathrm{\τ}},f_{\mathrm{\η}})\≈-\frac{4\mathrm{\π}(R_0-R_{ref})}{c}\sqrt{{(f_0+f_{\mathrm{\τ}})}^2-\frac{c^2{f}_{\mathrm{\η}}^2}{4V_r^2}}---\left(2\right)$

Wherein, R₀For target range to position, R_refFor reference distance, f_τFor orientation to the frequency of distance of unit, f₀Table Show that carrier frequency, c represent the light velocity, V_rRepresent radar speed.

Doppler centroid f_ηcExpression formula is as follows:

$f_{\mathrm{\η}c}=\frac{2V_r{\mathrm{sin\θ}}_{r,c}}{\mathrm{\λ}}---\left(3\right)$

Wherein, θ_r,cFor the angle of strabismus of beam center, λ is wavelength.f_ηcMake f_ηMore than actual value.As in figure 2 it is shown, its In (a) be the frequency spectrum before Stolt interpolation, (b) is the frequency spectrum after Stolt interpolation, and in the case of stravismus, Stolt maps Rear f_τ' relatively f_τThere is bigger displacement, and different to the value of the displacement of unit for different azimuth, there is oblique pull and twisted phenomena. Therefore each orientation f to unit is estimated by distance to spectral range_τIt is mapped to f_τ' minima on axle is as initial Value, and it is carried out quantization round:

$f_{\mathrm{\τ},min}^{\′}\[i\]=ceil\left(sqrt\right({\left(f_0+f_{\mathrm{\τ}}\[0\]\right)}^2-\frac{c^2{f}_{\mathrm{\η}}^2\[i\]}{4V_r^2})/(f_s/Nr\left)\right)---\left(4\right)$

Wherein, f_τ'_,min[i] represents the i-th orientation f to unit_τ' quantify the minima after rounding, f_τ[0] distance is represented Frequency initial value, f_η[i] represent i-th orientation to the orientation of unit to frequency, f_sRepresenting sample frequency, Nr is former The distance of beginning echo data S is to sampling number.

Step 2, with f_τ'_,minAs initial value, withFor frequency interval, the frequency of distance calculating each data point is reflected Penetrate f_τ' value.

I-th orientation maps f to the frequency of distance of unit_τ'_,iMay be defined as:

$f_{\mathrm{\τ},i}^{\′}=f_{\mathrm{\τ},min}^{\′}\[i\]\frac{f_s}{Nr}+\frac{(1:Nr)}{Nr}{f}_s---\left(5\right)$

Data after consistent compression store with the form of two-dimensional matrix, if the position coordinates of data point be (i, k), wherein I be orientation to coordinate, k be distance to coordinate, then formula (5) can be expressed as:

$f_{\mathrm{\τ},ik}^{\′}=f_{\mathrm{\τ},min}^{\′}\[i\]+\frac{k}{Nr}{f}_s---\left(6\right)$

Step 3, by Stolt mapping equation, calculate the f of each data point_τ' it is worth correspondence at f_τThe position of axle, and calculate Go out interpolation result.

Try to achieve f_τ'_,ikAfter, substituted in (1) formula:

$\sqrt{{(f_0+f_{\mathrm{\τ},ik})}^2-\frac{c^2{f}_{\mathrm{\η}}^2}{4V_r^2}}=f_0+f_{\mathrm{\τ},ik}^{\′}---\left(7\right)$

After equation converts, try to achieve it at f_τThe mapping value of axle:

$f_{\mathrm{\τ},ik}=\sqrt{{(f_0+f_{\mathrm{\τ},ik}^{\′})}^2+\frac{c^2{f}_{\mathrm{\η}}^2}{4V_r^2}}-f_0---\left(8\right)$

f_τ,ikAfter trying to achieve, it is carried out resampling in distance on frequency spectrum, in order to ensure that precision typically uses sinc interpolation to enter Row resampling.Each orientation maps f to the frequency of distance of unit_τ' original position the most different, this is done to revise Frequency spectrum, it is ensured that all spectrum components both fall within former support region, increases support region utilization rate.

Next linear phase is compensated and illustrate.

After revising Stolt interpolation, each orientation is to the f of unit_τ' original position be different, this will distance to After IFFT, data are carried out linear phase compensation so that each orientation is to the f of unit_τ' alignment.

Discrete Fourier transform character frequency shift property is:

$x\left(n\right)e^{-{\mathrm{j\ω}}_{0}n}=X\left(e^{j(\mathrm{\ω}+{\mathrm{\ω}}_0)}\right)---\left(9\right)$

Wherein, x (n) is time domain discrete sequence, X (e^jω) it is frequency spectrum corresponding for x (n), ω₀For spectrum offset amount, ω is Numeral angular frequency, ω and simulation angular frequency Ω and sample frequency f_sRelation be:

$\mathrm{\ω}=\frac{\mathrm{\Ω}}{f_s}---\left(10\right)$

Coordinate position be (i, data point k) in the side-play amount revising Stolt interpolation middle-range descriscent frequency domain is:

${\mathrm{\Ω}}_0=2{\mathrm{\πf}}_{\mathrm{\τ},min}^{\′}\[i\]\frac{f_s}{Nr}---\left(11\right)$

Formula (11) is substituted into formula (9) and formula (10), and this is equivalent to be multiplied by following phase place in distance to time domain:

$\exp (-j2\mathrm{\π}m\frac{f_{\mathrm{\τ},\min }^{\′}\[i\]}{Nr})---\left(12\right)$

It is thus desirable to distance after IFFT to data point S_ikFilling this phase place, expression formula is as follows:

$S_{ik}^{\′}=S_{ik}\exp \left(j2\mathrm{\π}m\frac{f_{\mathrm{\τ},\min }^{\′}\[i\]}{Nr}\right)---\left(13\right)$

Wherein, m is positive integer.

Owing to the phase multiplication of time domain is equivalent to do cyclic shift at frequency domain, after phase compensation, each orientation To the f of unit_τ' alignd, although 2-d spectrum has recovered oblique pull characteristic, but owing to cyclic shift carries out replicate, no Support region can be exceeded.

Finally carry out orientation to IFFT, then can obtain imaging results.

Effectiveness of the invention is further illustrated below by point target emulation experiment.

Software platform used by emulation experiment of the present invention is MATLAB.

In emulation experiment, the scattergram of point target is as shown in Figure 3.It is as shown in the table for radar parameter:

Fig. 4 (a) is the 2-d spectrum after consistent compression, and Fig. 4 (b) is frequency spectrum after revising Stolt interpolation, permissible Find out that the frequency spectrum after interpolation fall into support region originally substantially.Fig. 4 (c) is for carry out linear phase at range-Dopler domain 2-d spectrum after compensation, this step recovered spectral distortion and the oblique pull that traditional Stolt interpolation is brought, but due to time Territory is multiplied by linear phase and is equivalent to the cyclic shift at frequency domain, and therefore Fig. 4 (c) intermediate frequency spectrum has carried out replicate, without departing from Supporting domain.

Fig. 5 is the imaging results figure of emulation experiment.Fig. 6 (a) is the contour map of central point target, and 6 (b) is upper right The contour map of point target.Can be seen that the method that the present invention proposes can obtain good imaging in the case of large slanting view angle machine Result.

The above is only the some embodiments of the present invention, it is noted that for those skilled in the art For, under the premise without departing from the principles of the invention, it is also possible to make some improvement, these improvement should be regarded as the present invention's Protection domain.

Claims (8)

1. the improvement Omega-K formation method of a large slanting view angle machine SAR, it is characterised in that comprise the following steps:

Step 1: obtain raw radar data S；

Step 2: raw radar data S is successively carried out distance to fast Fourier transform FFT, distance to pulse compression, Distance is compressed with consistent to inverse discrete Fourier transform IFFT, motion compensation, Two-dimensional FFT；

Step 3: calculate spectrum offset amount f '_τ,min, revise Stolt interpolation；

Step 4: each data point is carried out distance to IFFT, transform data to range-Dopler domain, obtain data point S_ik；

Step 5: according to f '_τ,minCarry out linear phase compensation, complete distance to Spectrum Correction, obtain data point S '_ik；

Step 6: to data point S '_ikCarry out orientation to IFFT, obtain final imaging results.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 1, it is characterised in that In step 1, the size of raw radar data S is Na × Nr, wherein, Na be orientation to sampling number, Nr be distance to Sampling number.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 1 and 2, its feature exists In, in step 2, the data after consistent compression store with the form of two-dimensional matrix.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 1 and 2, its feature exists Following steps are specifically included in, step 3:

Step 3-1: computer azimuth is to frequency of distance f of unit_τIt is mapped to f ' by Stolt_τThe minima of axle, quantifies to take After whole, f ' is arrived in storage_τ,minIn；

Step 3-2: with f '_τ,minAs initial value, withFor frequency interval, the frequency of distance calculating each data point is reflected Penetrate f '_τValue；

Step 3-3: by Stolt mapping equation, calculate the f ' of each data point_τValue is corresponding at f_τThe position of axle, and count Calculate interpolation result.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 4, it is characterised in that In step 3-1, i-th orientation is to frequency of distance f of unit_τIt is mapped to f ' by Stolt_τAxle quantifies the minima after rounding Computational methods be:

$f_{\mathrm{\τ},min}^{\′}\[i\]=ceil\left(sqrt\right({\left(f_0+f_{\mathrm{\τ}}\[0\]\right)}^2-\frac{c^2{f}_{\mathrm{\η}}^2\[i\]}{4V_r^2})/(f_s/Nr\left)\right)$

Wherein, f '_τ,min[i] represents the i-th orientation f ' to unit_τQuantify the minima after rounding, f_τ[0] distance frequency is represented Rate initial value, f_η[i] represent i-th orientation to the orientation of unit to frequency, f_sRepresenting sample frequency, Nr is original time The distance of wave datum S is to sampling number, f₀Representing carrier frequency, c represents the light velocity, V_rRepresent radar speed.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 4, it is characterised in that In step 3-3, i-th orientation is to the f of unit_τ' it is worth correspondence at f_τThe position of axle is:

$f_{\mathrm{\τ},ik}=\sqrt{{(f_0+f_{\mathrm{\τ},ik}^{\′})}^2+\frac{c^2{f}_{\mathrm{\η}}^2}{4V_r^2}}-f_0$

Wherein, i be orientation to coordinate, k is that distance is to coordinate, f '_τ,ikRepresent that position coordinates is (i, the distance of data point k) Frequency is mapped in f '_τThe value of axle, f_τ,ikRepresent f '_τ,ikCorresponding at f_τThe position of axle.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 4, it is characterised in that Step 3-3 use sinc interpolation calculate interpolation result.

The improvement Omega-K formation method of large slanting view angle machine SAR the most according to claim 1, it is characterised in that In step 5, the data point after linear phase compensation is:

$S_{ik}^{\′}=S_{ik}\exp \left(j2\mathrm{\π}m\frac{f_{\mathrm{\τ},\min }^{\′}\[i\]}{Nr}\right)$

Wherein, f '_τ,min[i] represents the i-th orientation f ' to unit_τQuantifying the minima after rounding, Nr is original echo number According to the distance of S to sampling number, m is positive integer.